Research map / Continual learning / Online and replay-based
Crafting Your Evolving Dreams: Concept-Incremental Versatile Customization
Jiahua Dong, Wenqi Liang, Hongliu Li, Yang Cong, Duzhen Zhang, hanbin zhao, Henghui Ding, Yulun Zhang, Salman Khan, Fahad Shahbaz Khan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LCT module (ablation) | Base · Single-concept text-to-image (SDXL) | FIA: 9.44 -> 8.36 | Improved |
| add LCT module (ablation) | Base · Single-concept text-to-image (SDXL) | IA: 64.89 -> 65.18 | Improved |
| add LCT module (ablation) | Base · Single-concept text-to-image (SDXL) | TA: 80.28 -> 80.05 | No clear effect |
| add FGA module (ablation) | Base+LCT · Single-concept text-to-image (SDXL) | IA: 65.18 -> 65.76 | Improved |
| add FGA module (also cuts parameters) (ablation) | Base+LCT · Single-concept text-to-image (SDXL) | #Params: 6.43M -> 4.72M | Improved |
| add MKB module (ablation) | Base+LCT+FGA · Single-concept text-to-image (SDXL) | FIA: 8.06 -> 7.18 | Improved |
| add MKB module (fewer parameters) (ablation) | Base+LCT+FGA · Single-concept text-to-image (SDXL) | #Params: 4.72M -> 4.23M | Improved |
| add RGA module (ablation) | Base+LCT+FGA+MKB · Single-concept text-to-image (SDXL) | IA: 66.25 -> 72.84 | Improved |
| add RGA module (ablation) | Base+LCT+FGA+MKB · Single-concept text-to-image (SDXL) | FIA: 7.18 -> 3.84 | Improved |
| add RGA module (ablation) | Base+LCT+FGA+MKB · Single-concept text-to-3D (SDXL) | FIA: 5.54 -> 2.89 | Improved |
| use CCDM instead of CIDM (conference version) | CIDM · Single-concept text-to-image (SDXL) | FIA: 8.35 -> 3.84 | Improved |
| use CCDM instead of CIDM with fewer parameters | CIDM · Multi-concept text-to-image (SDXL) | IA: 72.12 -> 77.25 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CLoRA — C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models (NeurIPS 2026) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- How to Continually Adapt Text-to-Image Diffusion Models for Flexible Customization? (NeurIPS 2024 · similarity 0.93)
- ConceptGuard: Continual Personalized Text-to-Image Generation with Forgetting and Confusion Mitigation (CVPR 2025 · similarity 0.89)
- LoRACLR: Contrastive Adaptation for Customization of Diffusion Models (CVPR 2025 · similarity 0.87)
- ClassDiffusion: More Aligned Personalization Tuning with Explicit Class Guidance (ICLR 2025 · similarity 0.85)
- Orthogonal Adaptation for Modular Customization of Diffusion Models (CVPR 2024 · similarity 0.85)
- Enhancing Creative Generation on Stable Diffusion-based Models (CVPR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
